Building up time-consistency for risk measures and dynamic optimization
Published in European Journal of Operational Research, 2016
Recommended citation: De Lara, Michel, and Vincent Leclère. "Building up time-consistency for risk measures and dynamic optimization." European Journal of Operational Research 249.1 (2016): 177-187. https://doi.org/10.1016/j.ejor.2015.03.046
We study the management of a chain of dam hydroelectric production where we consider the expected gain stemming from the production as the criterion to maximize. However solving directly the problem by Dynamic Programming approach can be numerically impossible because of the so-called curse of dimension. Consequently we will use some decomposition-coordination method on this problem. However if decomposition-coordination methods are well known in a deterministic setting, few results are available in a stochastic setting. We will present a simple problem with three dams, that can be solved by dynamic programming, and the Dual Approximate Dynamic Programming (DADP) decomposition method we are using on this problem. As we have the exact solution of the problem, we can present a torough study of the numerical properties of DADP.